Psychotic symptoms are present in a significant subset of individuals with Bipolar Disorder (BD) and carry devastating personal and clinical implications. Most biomedical research on BD has ignored the variable presentation of psychosis possibly overlooking biologically significant heterogeneity in BD; such heterogeneity may cause inconsistencies in the literature by treating BD as a homogenous category 7,18. The expression of psychosis in some BD patients (BD-P) and absence in others (BD-NP) may indicate divergent disease processes of critical nosological and clinical relevance. PARDIP leverages a large sample of BD, a comprehensive battery, and sophisticated analytic tools to establish whether BD-P and BD-NP represent a difference in degree or a difference in kind. Long-term goals: This work will critically impact how BD is classified and studied, provide robust targets for effective future etiological studies, and clarify the utility of available biomarkers o major psychiatric disturbance. PARDIP represents a step toward mechanistically based classification of psychiatric disorders. Specific Aims: PARDIP will (i) identify the patterns of bo-cognitive disruptions which mark psychosis (BD-P`BD-NP) or mood instability in general (BD`healthy comparisons), (ii) explore how these biomarkers relate to one another and to other dimensions of psychopathology present in BD, and (iii) utilize latent class and cluster analyses of the multivariate dataset to verify taxonicity within BD with regard to psychosis and uncover latent psychiatric subgroups of interest for future genotyping and etiological research. Methods: The three-year PARDIP project will recruit 135 psychotic BD, 135 non-psychotic BD, and 135 psychiatrically healthy comparison subjects (all new recruits), administering a comprehensive battery focused on the psychosis and mania domains of psychopathology. We will obtain measures of neurophysiology, (smooth pursuit eye movements, antisaccades, auditory ERPs), cognition (cognitive battery, response inhibition, spatial working memory), neuroanatomy (structural magnetic resonance imaging [MRI]), emotional processing (ERPs to emotional pictures), intrinsic brain state (resting functional MRI connectivity), and circadian function (Actigraphy). We will compare biomarkers between BD-P, BD-NP, and H groups to determine which track with psychosis and which track with affective disturbance. We will identify common sources of variance among measures with joint-ICA and PCA approaches, and examine how biomarkers and biomarker composites relate to other aspects of clinical heterogeneity. Taxometric procedures (MAXCOV- HITMAX and its multivariate extension MAXEIG-HITMAX and k-means clustering) will be carried out with the multivariate dataset to empirically identify distinct subgroups of subjects. PARDIP will be conducted by 4 experienced research groups (across 3 collection sites) with a long history of close and productive collaboration.